Chapter 7 ■ Section 4 Constrained Optimization: The Method of Lagrange Multipliers 557
APPLICATIONS TO ECONOMICS
MAXIMIZATION OF UTILITY
In the next two examples, the method of Lagrange multipliers is used to solve con-
strained optimization problems from economics.
A utility function U(x, y) measures the total satisfaction or utility a consumer receives
from having x units of one particular commodity and y units of another. The problem is
to determine how many units of each commodity the consumer should buy to maximize
utility while staying within a fixed budget. The application of the method of Lagrange
multipliers to the utility problem is illustrated in the next example.
EXAMPLE 4.4
A consumer has $600 to spend on two commodities, the first of which costs $20 per
unit and the second $30 per unit. Suppose that the utility derived by the consumer
from x units of the first commodity and y units of the second commodity is given by
the Cobb-Douglas utility function U(x, y) 5 10x0.6y0.4. How many units of each
commodity should the consumer buy to maximize utility?
Solution
The total cost of buying x units of the first commodity at $20 per unit and y units of
the second at $30 per unit is 20x 1 30y. Since the consumer has only $600 to spend,
the goal is to maximize utility U(x, y) subject to the budgetary constraint that 20x 1
30y 5 600.
The three Lagrange equations are
6x20.4y0.4 5 20l 4x0.6y20.6 5 30l and 20x 1 30y 5 600
From the first two equations you get
6x20.04y0.4 4x0.6y20.6
5 5l
20 30
9x20.4y0.4 5 4x0.6y20.6
4
9y 5 4x or y5 x
9
Substituting this into the third equation, you get
20x 1 30 194 x2 5 600
from which it follows that
4
x 5 18 and y 5 (18) 5 8
9
That is, to maximize utility, the consumer should buy 18 units of the first commodity
and 8 units of the second.
 558 Chapter 7 Calculus of Several Variables

Optimal indifference curve: U(x, y) = C

20

Budget line: 20x + 30y = 600

(18, 8)

x
30

FIGURE 7.22 Budgetary constraint and optimal indifference curve.

Recall from Section 1 that the level curves of a utility function are known as
indifference curves. A graph showing the relationship between the optimal indiffer-
ence curve U(x, y) 5 C, where C 5 U(18, 8) and the budgetary constraint 20x 1 30y
5 600, is sketched in Figure 7.22.

ALLOCATION OF RESOURCES An important class of problems in business and economics involves determining an
optimal allocation of resources subject to a constraint on those resources. Here is an
example in which sales are maximized subject to a budgetary constraint.

EXAMPLE 4.5
An editor has been allotted $60,000 to spend on the development and promotion of
a new book. It is estimated that if x thousand dollars is spent on development and y
thousand on promotion, approximately f(x, y) 5 20x3/2y copies of the book will be
sold. How much money should the editor allocate to development and how much to
promotion in order to maximize sales?

Solution
The goal is to maximize the function f(x, y) 5 20x3/2y subject to the constraint
g(x, y) 5 60, where g(x, y) 5 x 1 y. The corresponding Lagrange equations are
30x1/2y 5 l 20x3/2 5 l and x 1 y 5 60
Chapter 7 ■ Section 4 Constrained Optimization: The Method of Lagrange Multipliers 559

From the first two equations you get

30x1/2y 5 20x3/2
Since the maximum value of f clearly does not occur when x 5 0, you may assume
that x Þ 0 and divide both sides of this equation by 30x1/2 to get
2
y5 x
3
Substituting this expression into the third equation, you get
2 5
x 1 x 5 60 or x 5 60
3 3
from which it follows that
2
x 5 36 and y 5 (36) 5 24
3
That is, to maximize sales, the editor should spend $36,000 on development and
$24,000 on promotion. If this is done, approximately f(36, 24) 5 103,680 copies of
the book will be sold.
A graph showing the relationship between the budgetary constraint and the level
curve for optimal sales is shown in Figure 7.23.

Optimal sales level: f(x, y) = 103,680

60

(36, 24)
Budget line: x + y = 60

x
60

FIGURE 7.23 Budgetary constraint and optimal sales level.

 560 Chapter 7 Calculus of Several Variables

THE SIGNIFICANCE OF THE You can solve most constrained optimization problems by the method of Lagrange
LAGRANGE MULTIPLIER l multipliers without actually obtaining a numerical value for the Lagrange multiplier
l. In some problems, however, you may want to compute l. This is because l has
the following useful interpretation.

The Lagrange Multiplier ■ Suppose M is the maximum (or mini-

mum) value of f(x, y), subject to the constraint g(x, y) 5 k. The Lagrange mul-
tiplier l is the rate of change of M with respect to k. That is,
dM
l5
dk
Hence,
l . change in M resulting from a 1-unit increase in k

EXAMPLE 4.6
Suppose the editor in Example 4.5 is allotted $61,000 instead of $60,000 to spend on
the development and promotion of the new book. Estimate how the additional $1,000
will affect the maximum sales level.

Solution
In Example 4.5, you solved the three Lagrange equations
30x1/2y 5 l 20x3/2 5 l and x 1 y 5 60
to find that the maximum value M of f(x, y) subject to the constraint x 1 y 5 60
occurred when x 5 36 and y 5 24. To find l, substitute these values of x and y into
the first or second Lagrange equation. Using the second equation, you get
l 5 20(36)3/2 5 4,320
which means that maximal sales will increase by approximately 4,320 copies (from
103,680 to 108,000) if the budget is increased from $60,000 to $61,000.

WHY THE METHOD Although a rigorous explanation of why the method of Lagrange multipliers works
OF LAGRANGE involves advanced ideas beyond the scope of this text, there is a rather simple geo-
MULTIPLIERS WORKS metric argument that you should find convincing. This argument depends on the fact
that for the level curve F(x, y) 5 C, the slope is given by
dy Fx
52
dx Fy
Chapter 7 ■ Section 4 Constrained Optimization: The Method of Lagrange Multipliers 561

This result is true for any level curve of a function F whose partial derivatives exist
(provided Fy Þ 0). An example justifying the formula is outlined in Problem 48.
Now, consider the constrained optimization problem:
Maximize f(x, y) subject to g(x, y) 5 k
Geometrically, this means you must find the highest level curve of f that intersects the
constraint curve g(x, y) 5 k. As Figure 7.24 suggests, the critical intersection will occur
at a point where the constraint curve is tangent to a level curve; that is, where the slope
of the constraint curve g(x, y) 5 k is equal to the slope of a level curve f(x, y) 5 C.

y
Direction in which C increases

Point of tangency

Highest level curve,

f(x, y) = C, intersecting
the constraint

Constraint curve: g(x, y) = k

FIGURE 7.24 Increasing level curves and the constraint curve.

According to the formula stated at the beginning of this discussion, you have
Slope of constraint curve 5 Slope of level curve
gx fx
2 52
gy fy

or, equivalently,
fx fy
5
gx gy

If you let l denote this common ratio, then

fx fy
5l and 5l
gx gy